Last week I highlighted potential value in siding with Donald Trump in Iowa at a healthy 5/4. That price has long gone, and the best currently available is 4/6. My previously advised 13/8 about Trump in Florida has similarly shortened to a best price of 11/10. So let’s see if we can identify another State which is currently mispriced. To do this, I have decided to enlist the help of my colleague Prof David Paton, of Nottingham University Business School, to this time look at North Carolina. In 2012, Romney won the State back for the Republicans with a margin of 2%. Demographic changes such as increasing numbers of Hispanics, however, mean that it is a key target this time for the Democrats. And here the polls and forecasters agree. The Real Clear Politics average has Clinton 2.6% up on Trump (2.7% on HuffPost Pollster), whilst New York Times election analyst, Nate Cohn, has been estimating a huge 6% lead for Clinton. Over at Fivethirtyeight, Nate Silver’s polls-based forecast currently rates Hillary’s chances of winning the Tar Heel State at between 57% and 58.1%.

Our view is that the evidence we have from the early voting has not been factored in sufficiently. Of all the States for which we have early voting information, North Carolina provides the best data. We can track votes cast broken down by party registration and a range of other demographics. Even better, we can track how 2016 is shaping up compared to the equivalent dates before the 2012 election. Our analysis of that data throws up some striking findings. As of the most recent early votes cast, the vote share of registered Republicans is virtually the same as the equivalent date in 2012. In contrast, the Democrat share has gone down by 11% from 49.1% on Tuesday 30th October 2012 to 43.7% today. Nate Cohn has an explanation for such a drop that could mean it is not such good news for Trump. Specifically that Democrat voter registration has dropped since 2012, by 8.3%, but that the voters who remain (notably college-educated white women) trend Democrat to such an extent that it outweighs this plunge. A big assumption. Given that the Democrats need to come from behind to win this State, the raw numbers do not look like great news from their point of view.

But another trend is even more striking. The share of early voting amongst black voters has tumbled by 21.5% compared to the equivalent date in 2012, whilst early voting overall is up 8.4%. The early voting share for whites has gone up over 7%. There is very little change in the gender mix of early voting. We know that black voters tend to fall overwhelmingly into the Democrat column, so the huge drop in turnout in that group is potentially devastating for Hillary Clinton.

Does that mean Trump is a sure-fire winner of North Carolina? Of course not – perhaps Republican-registered women are indeed actually voting for Hillary in large numbers whilst every day brings the prospect of another video which could shake up the race again and hit Trump in every state. But on the basis of early voting, winning North Carolina is going to be an uphill battle for Secretary Clinton and the current price of 13/8 about Trump looks very tempting.

So what about Iowa? Although the early voting figures in the Hawkeye State also gives some positive indications for Trump, the situation relative to 2012 is not as good as in North Carolina. As a result, the current Trump price of 4/6 in Iowa looks distinctly mean compared to the 7/4 available about a Republican win in North Carolina. It is quite possible that Trump will win both Iowa and North Carolina. He might lose both States. He could also win North Carolina and lose Iowa. But our analysis of the early voting suggests at the moment a scenario in which Trump wins Iowa but loses North Carolina carries with it a probability very much smaller than is implied in the odds.

Follow Leighton on Twitter @leightonvw

Back Trump to win North Carolina - 1pt @ 13/8